Understanding an R corpus

1. Understanding an R corpus

Hello again, and welcome to chapter 2. We will soon build off of the regular expression and tokenization lessons from chapter 1, but first, lets explore collections of texts, which we will call a Corpus.

2. Corpora

From the R documentation, corpora are collections of documents containing natural language text. The most common representation of this is from the tm package, and is called a Corpus. "Corpora" is the plural form of "corpus." More specifically though, a VCorpus, or volatile corpus, is usually used to host both the text,as well as the metadata about the collection of text we are using.

3. Contents of a VCorpus: metadata

The tm package contains an example dataset for us to use, called acq, which contains 50 articles from the Reuters dataset. The data is not that important here, but understanding the VCorpus structure is. For the first article, we access the meta data by calling the meta item. There is data about the author, a time stamp, heading, and much more. All of this data might be useful to our analysis going forward and it is important that we know how to access it.

4. Contents of a VCorpus: metadata

Furthermore, the individual objects within meta can also be accessed. If you have not worked with nested objects before, pay attention to this structure. Here we accessed the meta item of the first article and then the character value for the place where the article originated, which has a value of USA.

5. Contents of a VCorpus: content

We can also see that the VCorpus contains the actual text of each document by accessing the content item. Keeping the contents of a VCorpus as is allows us to use several analysis functions from the tm package. However, to do other analysis, we need to have a tidier version of the data.

6. Tidying a corpus

In order to get the data into a table format, where each observation is represented by a row, and each variable is a column, we can use the tidy() function on a VCorpus. This will grab the values of each article, for both the metadata and the content, and convert it into a tibble. We have already been working with tibbles when we tokenized and removed stopwords in the previous chapter.

7. Creating a corpus

Of course there is the flip side to this operation. Perhaps you are interested in using some analysis functions from the tm package, but you only have a tibble. We can convert the text from a tibble into the format of a VCorpus. We use the VCorpus() function, along with the VectorSource() function, to create a corpus. This basically tells R to create a corpus by using a vector of text. However, this only captures the text. To add the meta information from the tidy data, we use the meta function, which adds a column to the metadata data frame attached to the corpus.

8. Let's see this in action.

Let's practice going back and fourth between corpus objects and tibbles.

This exercise is part of the course

Introduction to Natural Language Processing in R

IntermediateSkill Level

5.0+

Start Course for Free

Chapter 1 of Introduction to Natural Langauge Processing prepares you for running your first analysis on text. You will explore regular expressions and tokenization, two of the most common components of most analysis tasks. With regular expressions, you can search for any pattern you can think of, and with tokenization, you can prepare and clean text for more sophisticated analysis. This chapter is necessary for tackling the techniques we will learn in the remaining chapters of this course.

Exercise 1: Regular expression basics Exercise 2: Practicing syntax with grep Exercise 3: Exploring regular expression functions.Exercise 4: Tokenization Exercise 5: tidytext functions Exercise 6: Tokenization: sentences Exercise 7: Text cleaning basics Exercise 8: Text preprocessing: remove stop words Exercise 9: Text preprocessing: Stemming

In this chapter, you will learn the most common and studied ways to analyze text. You will look at creating a text corpus, expanding a bag-of-words representation into a TFIDF matrix, and use cosine-similarity metrics to determine how similar two pieces of text are to each other. You build on your foundations for practicing NLP before you dive into applications of NLP in chapters 3 and 4.

Exercise 1: Understanding an R corpus

Current Exercise

Exercise 2: Explore an R corpus Exercise 3: Creating a tibble from a corpus Exercise 4: Creating a corpus Exercise 5: The bag-of-words representation Exercise 6: Practice BoW Exercise 7: BoW Example Exercise 8: Sparse matrices Exercise 9: The TFIDF Exercise 10: Manual calculations Exercise 11: TFIDF Practice Exercise 12: Cosine Similarity Exercise 13: An example of failing at text analysis Exercise 14: Cosine similarity example

Chapter 3 focuses on two common text analysis approaches, classification modeling, and topic modeling. If you are working on text analysis projects, you will inevitably use one or both of these methods. This chapter teaches you how to perform both techniques and provides insight into how to approach these techniques from a practical point of you.

Exercise 1: Preparing text for modeling Exercise 2: Data preparation Exercise 3: Removing sparse terms Exercise 4: Classification modeling Exercise 5: Classification modeling example Exercise 6: Confusion matrices Exercise 7: TFIDF tibble vs dtm Exercise 8: Introduction to topic modeling Exercise 9: LDA practice Exercise 10: Assigning topics to documents Exercise 11: LDA in practice Exercise 12: Testing perplexity Exercise 13: Reviewing LDA results

In chapter 4 we cover two staples of natural language processing, sentiment analysis, and word embeddings. These are two analysis techniques that are a must for anyone learning the fundamentals of text analysis. Furthermore, you will briefly learn about BERT, part-of-speech tagging, and named entity recognition. Almost 15 different analysis techniques were covered in this course, so chapter 4 ends by recapping all of the great techniques you will learn about in this course.

Exercise 1: Sentiment analysis Exercise 2: tidytext lexicons Exercise 3: Sentiment scores Exercise 4: Sentiment and emotion Exercise 5: Word embeddings Exercise 6: h2o practice Exercise 7: word2vec Exercise 8: Additional NLP analysis Exercise 9: Reviewing methods #1 Exercise 10: Review methods #2 Exercise 11: Conclusion